Add parallel LLM racing implementation to 45-llm-hedge.py

examples/foundational/45-llm-hedge.py

@@ -0,0 +1,192 @@
+#
+# Copyright (c) 2024–2025, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import os
+from typing import override
+
+from dotenv import load_dotenv
+from loguru import logger
+from openai.types.chat import ChatCompletionMessageParam
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import Frame, LLMTextFrame
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.openai_llm_context import (
+    OpenAILLMContext,
+    OpenAILLMContextFrame,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.network.fastapi_websocket import FastAPIWebsocketParams
+from pipecat.transports.services.daily import DailyParams
+
+load_dotenv(override=True)
+
+# We store functions so objects (e.g. SileroVADAnalyzer) don't get
+# instantiated. The function will be called when the desired transport gets
+# selected.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+        vad_analyzer=SileroVADAnalyzer(),
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+        vad_analyzer=SileroVADAnalyzer(),
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+        vad_analyzer=SileroVADAnalyzer(),
+    ),
+}
+
+
+class LLMRaceProcessor(FrameProcessor):
+    """Manages racing between two LLMs - only allows frames from the first LLM to respond."""
+
+    # Class variables to share state between instances (using public names)
+    winning_llm_name = None
+    response_started = False
+
+    def __init__(self) -> None:
+        super().__init__()
+        self._current_llm_name = None
+
+    def set_llm_name(self, name: str):
+        """Set the name of the LLM this processor instance is handling."""
+        self._current_llm_name = name
+
+    @override
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        # Always call super first to handle StartFrame and other system frames
+        await super().process_frame(frame, direction)
+
+        if isinstance(frame, LLMTextFrame):
+            if not LLMRaceProcessor.response_started:
+                # First response wins the race
+                LLMRaceProcessor.winning_llm_name = self._current_llm_name
+                LLMRaceProcessor.response_started = True
+                logger.info(
+                    f"🏆 [LLM_RACE] {self._current_llm_name} wins the race! Text: '{frame.text}'"
+                )
+                await self.push_frame(frame, direction)
+            elif LLMRaceProcessor.winning_llm_name == self._current_llm_name:
+                # Continue allowing frames from winning LLM
+                logger.info(f"✅ [LLM_RACE] {self._current_llm_name} continuing: '{frame.text}'")
+                await self.push_frame(frame, direction)
+            else:
+                # Drop frames from losing LLM
+                logger.info(
+                    f"❌ [LLM_RACE] Dropping '{frame.text}' from losing LLM: {self._current_llm_name}"
+                )
+        else:
+            # Pass through all non-LLM frames (including system frames)
+            await self.push_frame(frame, direction)
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot with parallel LLM racing")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY", ""))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY", ""),
+        voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+    )
+
+    # Create two LLM instances for racing
+    llm1 = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    llm2 = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    messages: list[ChatCompletionMessageParam] = [
+        {
+            "role": "system",
+            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio so don't include special characters in your answers. Respond to what the user said in a creative and helpful way.",
+        },
+    ]
+
+    # Create shared context for both LLMs
+    context = OpenAILLMContext(messages)
+    context_aggregator = llm1.create_context_aggregator(context)
+
+    # Make sure both LLMs share the same context - they should both process context frames
+    # In a ParallelPipeline, the context frames will be duplicated to both branches
+
+    # Create separate race processors for each LLM to track which one responds first
+    race_processor1 = LLMRaceProcessor()
+    race_processor1.set_llm_name("LLM1")
+
+    race_processor2 = LLMRaceProcessor()
+    race_processor2.set_llm_name("LLM2")
+
+    # Create parallel LLM branches using ParallelPipeline
+    parallel_llms = ParallelPipeline(
+        [llm1, race_processor1],  # Branch 1: LLM1 -> race processor 1
+        [llm2, race_processor2],  # Branch 2: LLM2 -> race processor 2
+    )
+
+    # Simple pipeline with parallel LLM processing
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # Speech to text
+            context_aggregator.user(),  # User responses (creates context frames for LLMs)
+            parallel_llms,  # Parallel LLM processing
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            context_aggregator.assistant(),  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        # observers=[debug_observer, llm_observer],
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Use a simpler approach - add message to context and push a context frame
+        messages.append({"role": "system", "content": "Please introduce yourself to the user."})
+        # Create a new context with the updated messages
+        updated_context = OpenAILLMContext(messages)
+        await task.queue_frames([OpenAILLMContextFrame(context=updated_context)])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()